Research Headlines – Tracking and behavioural analysis in public places

An EU-funded project is developing a blueprint for the design of low-cost approach to behaviour analysis in public places, while respecting individuals’ right to privacy. Such systems could be used for gauging crowd size and monitoring crowd flows for public safety and commercial purposes, for example.

© tashatuvango #101075578, 2019 source:

The evolution of information and communication technologies has led to the development of applications in which sensors are embedded in physical objects and connected to personal devices – a connected network called the Internet of Things (IoT).

The move to IoT makes it possible to analyse behaviour in public spaces. Current methods of analysis rely on visual systems or interactions between dedicated hardware and personal devices. However, such approaches give rise to significant privacy issues.

The EU-funded PATH project is defining a way to design detection, tracking and behaviour analysis systems which are simple, accurate and minimally intrusive, and which never identify individuals.

Such systems would not require the installation of active devices. Rather, the required capabilities would be integrated into existing IoT infrastructures in a given environment, such as a shopping centre or event venue.

The systems covered by the research are referred to as ‘passive’. In so-called ‘active’ systems, signals are exchanged directly with a target person or object. Passive systems reflect signals off the target to provide an estimation of its location. Both burglar alarms and radar operate in this way.

‘The solution proposed in PATH relies on sensor radar networks, where a signal transmitted in the area being monitored is reflected off people and things,’ says project coordinator Stefania Bartoletti of the University of Ferrara in Italy.

‘A network of sensors picks up the reflections and detects the presence of targets and events related to their behaviour through low-complexity signal processing and inferential techniques. PATH takes such ideas even further and investigates the possibility of eliminating any specialised transmission hardware and relying on signals that are there anyway, such as those from Wi-Fi or cellular systems,’ Bartoletti explains.

Recognising behavioural patterns

PATH has devised methods for identifying and making best use of suitable IoT networks for the deployment of passive systems. Based on the properties of wireless networks, the project has also developed techniques for processing and analysing signals and drawing inferences regarding target behaviour.

An approach to recognition of behavioural patterns has been tested in two demonstration activities. These covered key areas in which the outcomes of PATH could be implemented: monitoring and tracking of crowd flows in places like shopping centres and airports to determine which retail outlets are the most popular or where resources would be best allocated; and crowd counting at events where there are no turnstiles for public safety and measuring an event’s success.

Smart applications

Following the demonstrations, target localisation algorithms have been further developed to validate the approach under the supervision of a wireless sensor network company. The link lays the ground for application in industrial systems and a possible future spin-off.

‘In the coming years, the IoT promises to trigger a new industrial revolution – known as Industry 4.0 or Smart Factory – encompassing many technologies, starting with advanced manufacturing and moving into other segments, including smart logistics, smart cities and smart buildings,’ says Bartoletti. ‘In this context, the monitoring and tracking of people and things through the exploitation of diverse technologies and infrastructures already in place will enable development of a number of new applications and services.’

The project has received funding through the EU’s Marie Skłodowska-Curie Individual Fellowships programme and is being carried out in part with the Wireless Information and Network Sciences Laboratory of the Massachusetts Institute of Technology.

Leave a Reply

Your email address will not be published. Required fields are marked *